Keywords: biomonitoring, discriminant analysis, Hotelling T² test, multivariate analysis, principal component analysis, PCA, pollution prevention, environmental pollution, trace metals, mollusc, water pollution, Patella caerulea, Monodonta turbinata, statistical methods
Multivariate statistical methods applied to biomonitoring studies
This work describes the use of some multivariate statistical methods for retrieving the information present in a typical dataset coming from a biomonitoring study applied to trace metals measured on Patella caerulea and Monodonta turbinata samples. These samples were collected at five stations in Linosa Island in the Sicilian Channel (Mediterranean Sea) and the multivariate statistical methods applied were Principal Component Analysis (PCA), Hotelling T² test and Discriminant Analysis by means of Mahalanobis Distance (DAMD). Using PCA it is possible to have in a single and fast step, a direct insight into the structure of data, and thus obtain peculiar information that allows the different accumulation characteristics of the two types of mollusc to be seen. PCA also gives information about the metals that cause the peculiar accumulation characteristics of each mollusc. Hotelling's T² and DAMD are other important multivariate methods that confirm the results of PCA and also give complementary information for the interpretation of experimental data.